Collective Behaviour And Group Identification For Active Particles

Thanks to the development of technology in recent decades,more and more attention has been paid to the research of various complex systems,one of which is to study the behaviors of human beings and other movable creatures on spatial motion.Like collective foraging,defense,migration of animals and crowd behavior of human beings,a variety of collective behaviors emerge in biological groups under different situations.Understanding the occurrence mechanism for these phenomena on basic level makes us not only be able to take an ample look into living beings,but also be skilled at controlling collective behaviors.Researchers have already accumulated rich knowledge on both individual behaviors and collective behaviors in living beings' spatial motion from experimental observation and data analysis for different biological species.By establishing corresponding models,they can reproduce some of the phenomena in real groups.In this thesis,we studied the response behavior of pedestrian crowds to a special environmental constraint based on the anticipatory repulsion model according to Ioannis Karamouzas et al.,expecting to provide inspiration for the study of relationship between crowd and environment.Also,the crowd evacuation in the case of a single bottleneck exit is simulated,and some optimization methods are studied.Another trend in complex system research is the fact that data-driven research has become popular.The compressed sensing(CS)method is one of many new data analysis methods used by researchers.When the true solution for an underdetermined linear inverse problem is sparse,CS method could find it exactly only use little constraint equations,which means the time series for variables are not necessary to be long.On the spatial motion of human crowd,we look the whole crowd as an interacton network and uncover the network structure from individual motion data by using CS method.Some matters needing attention for CS application are discussed.The structure and innovation of this thesis are as follows.Chapter 1: In this chapter,several basics for active particle systems are reviewed.Firstly,we introduce three methods in describing the spatial motion of active paticles,including hydrodynamics,cellular automaton and social forces.Next,some common factors,such as leadership,positive feedback and negtive feedback,in collective behaviors are listed.At last,we show the roles of some environmetal factors,such as exits and obstacles,according to some researchers.Chapter 2: In this chapter,we simulate the pedestrian crowd motion in a special asymmetric annular channel based on Ioannis Karamouzas' anticipatory repulsion model.The results show that large crowds are sensitive to channel asymmetry while small crowd do not show a clear response.For large crowd,there exists a threshold preferred speed vpc,determining whether the whole crowd would rotate clockwise or anticlockwise at last.Chapter 3: In this chapter,we simulate the escaping crowd at a bottleneck exit base on Ioannis Karamouzas' anticipatory repulsion model.We argue that random control of appropriate intensity makes the crowd evacuate more smoothly.According to previous research,setting up circular obstacles may be helpful in evacuation.Therefore,we examine the effects of sizes and locations of circular obstacles on crowd evacuation.Chapter 4: In this chapter,we give compressed sensing method an intuitive explanation and apply it to interaction network detection for human crowd.Motion data is obtained from simulated crowd.we use this data to infer the network structure and discuss the factors that may affect the accuracy of inference.Considering we may not know in advance the sparsity of the solution in some practical problems and thus not know whether the length of time series used in the algorithm is sufficient to obtain a convincible result,we give a way to determine if the length of the time series is sufficient.Chapter 5: In this chapter,the contents of the researches listed in this thesis are briefly summarized.We also draw a prospect for possible follow-up works.